voiceio

Voice dictation for Linux that learns you. Local, private, open source, and it fine-tunes itself on your own voice.

Try it in your browser at voiceio.dev: no install, the model runs in the tab.

• Types anywhere. Press your hotkey, speak, and the words stream into the focused app with a live underlined preview (IBus), on Wayland and X11: GNOME, KDE, Hyprland, sway, i3.
• Runs on your machine. Whisper, or NVIDIA's Parakeet (text lands ~0.7 s after you stop talking instead of ~4.4 s, with every name you use as a hotword), decodes your speech locally. No account, no server, no telemetry; logs never contain your words.
• Learns your words. Keep your recordings (opt-in, on your disk) and voiceio fine-tunes its model on them while the machine is idle. It switches only when the new model wins on clips it never trained on. On the author's voice: 12.1% → 4.9% word errors after one night on a laptop CPU.

Install

With a coding agent. Paste this into Claude Code, Codex or any agent with a shell:

Install voiceio on this machine following https://voiceio.dev/llms.txt. Ask me which hotkey I want and whether to keep my recordings on this computer so voiceio can learn my voice. Finish with voiceio doctor and fix whatever it reports.

By hand:

sudo apt install pipx build-essential python3-dev portaudio19-dev ibus gir1.2-ibus-1.0 python3-gi   # Debian/Ubuntu
sudo dnf install pipx gcc gcc-c++ make python3-devel portaudio-devel ibus ibus-libs python3-gobject   # Fedora
sudo pacman -S python-pipx base-devel portaudio ibus python-gobject                                   # Arch
sudo emerge --ask media-libs/portaudio app-i18n/ibus dev-python/pygobject                             # Gentoo

pipx install 'python-voiceio[desktop]'      # or: uv tool install 'python-voiceio[desktop]'
voiceio setup        # model, hotkey, autostart

uvx is fine for a quick try, but run voiceio setup from a pipx or uv tool install: uvx runs from uv's cache, which uv deletes on uv cache clean, and the autostart service would point there.

Or as a Debian/Ubuntu package (new; bundles everything, nothing to compile):

# Debian / Ubuntu: download voiceio_*.deb from the latest release, then
sudo apt install ./voiceio_*.deb

voiceio setup

Then press your hotkey, speak, and press it again (or hold it while you speak). voiceio doctor shows what works, voiceio doctor --fix repairs what it can, and voiceio doctor --report shares an anonymous install report on the feedback board (it shows you the text and asks first). Doctor offers one when a check fails, and so does the tray's "Report a problem...". A report names the last error only by its type and file:line, never its message. NixOS, other extras and installing from source: docs/linux.md. Agent runbook: INSTALL.md.

Make it yours

voiceio vocab add Kalshi              # a name, recognized from the next dictation
voiceio corrections add kalchi Kalshi # fix whatever still comes out wrong
voiceio learn schedule on             # learn from your recordings while the machine is idle
voiceio learn status                  # what it learned and how each model scores

Scheduled learning runs daily or weekly at a time you pick, on mains power and only while you're away from the computer: it stops within seconds when you come back and resumes later where it was. You get a notification with a button to stop it. It never switches models unless the new one wins, and voiceio learn rollback undoes a switch. Details: docs/learning.md.

Let your agents talk to you

voiceio say --from build-bot "tests are green"     # chime, notification, then it's read aloud
voiceio say --from deploy --urgent "I need your approval"

Running several coding agents? Have each one call voiceio say when it needs you, so you hear which one. It never talks over you: while you dictate, announcements wait (speech into an open mic would get typed). They are read in your [tts] voice, offline.

Configure

Everything lives in ~/.config/voiceio/config.toml; config.example.toml documents every option. voiceio config set section.key value edits it in place.

Docs

Learning your voice	Vocabulary, corrections, labeling, fine-tuning, evaluation, schedule
Choosing a model	Whisper sizes, Parakeet, whisper.cpp servers, licenses
Commands	Every voiceio command
Your data	What is stored where, and what is off by default
Phone dictation	Run voiceio headless on a home server
Linux support and troubleshooting	Desktops, backends, NixOS, fixes
Comparisons	voiceio vs Voxtype, Wispr Flow, Superwhisper, Dragon and others

Roadmap

• Parakeet by default for everyone: it needs our sherpa-onnx fix (#4084) in an upstream release; until then voiceio models sherpa && voiceio models use parakeet-tdt-0.6b-v2 installs a patched build (why).
• GPU on AMD and Intel: let setup run a Vulkan whisper.cpp server for you.
• Typing without IBus on GNOME and KDE: a libei backend.
• Distro packages: AUR (packaged, not yet published) and .rpm.
• One model everywhere: sync your fine-tune across your machines.

Contributing

Open an issue before a large PR. CONTRIBUTING.md covers the architecture, conventions and the reasoning behind the model choice. Install reports and ideas are welcome on the feedback board.

How it's built

I write voiceio with AI coding assistants, mostly Claude Code. I decide what goes in and I use it every day, which is how most of its bugs get found.

Credits

• Speech recognition: faster-whisper and CTranslate2, with OpenAI's Whisper models; sherpa-onnx with NVIDIA's Parakeet-TDT. The in-browser demo runs Moonshine through transformers.js.
• Ideas borrowed with thanks: modifier release, paste keys and media pausing from Voxtype; the landing page's look from Romario Kavin's KOLlateral.

License

MIT. Model weights are not part of voiceio and carry their own licenses (see docs/models.md).